What is ResourceRouter?

And why can't the WPD give us good Shotspotter numbers?

Sep 12, 2024

This post was written for you, future reader, from the past (It’s Sunday Sept. 8 as I’m writing this). Trippy! By the time you’re reading this I’ll be in London, I’m pretty sure. And it will be the day after Katie’s birthday, which is why we’re traveling. How better to spend a 9/11 birthday than on a different continent all together! Everyone wish her a happy birthday! Any and all tips received for this post will go to lavishing her with London’s best Indian food.

venmo a tip :-)

Today’s post is a short and sweet examination of Shotspotter—what we know about it and moreso what we don’t.

Please subscribe! Here’s a little deal to sweeten the pot—the “Bill’s on vacation” special, which will end when I get back on the 19th.

Get 25% off for 1 year

And help spread the word!

What is ResourceRouter, and why can’t the WPD give us good numbers?

Worcester police finally provided a report on Shotspotter that Councilor Thu Nguyen asked for in May. The report was submitted at the last meeting in August, but held by Councilor King. It was on the agenda for Tuesday’s meeting, but I don’t know if there was any discussion (remember, I’m on vacation).

At just four pages, it’s lacking to say the least. This is a technology we spend almost $600,000 a year on, and we can only pry a few cold pages from the police department regarding it. But the absence of information leaves us with some interesting questions. The first one:

What is ResourceRouter?

From a report submitted to the city council in 2021.

When the council approved this “crime forecasting” tool in 2021, it had a different name: ShotSpotter Connect. At that point in time, its purpose was clearly stated—not so much the case anymore, as we’ll get to.

While it proved somewhat controversial at the time, the Worcester city council is not in the business of denying the cops a thing the cops want. They got their AI crime forecasting tool provided by the for-profit gunshot detection tech company.

Since then, we’ve received very little from the police about what this tool is or how they use it. The report provided by Saucier includes three terse paragraphs on the matter, and it’s the most substantive information we’ve gotten so far.

So let’s go through it piece by piece with some analysis.

ResourceRouter is a deployment tool based on the past five years of historical crime data from the public, not officer-initiated calls, and gives officers locations where their presence might prevent crime.

Not mentioned is that ResourceRouter is a repackaging of PredPol and HunchLab, two pieces of problematic software SoundThinking acquired when it bought the companies.

In April 2023, both the company and the service rebranded: Connect to ResourceRouter and the company to SoundThinking. The rebrand just so happened to come after Brandon Johnson won his mayoral election in Chicago, partially on a promise to part ways with ShotSpotter. The day after Johnson’s election, ShotSpotter stock fell 25 percent, to less than $28 a share. A few days later the company issued a press release about the name change and a suite of new technologies “for safer neighborhoods,” including ResourceRouter.

If the hope was to fortify its market share by distancing itself from negative press, it didn’t work...

Yikes! Down by half over five years. That’s not good.

In 2023, ShotSpotter acquired PredPol. It was the latest in a string of acquisitions, starting in 2018 with HunchLab (a PredPol competitor), then CrimeCenter in 2020, and Forensic Logic in 2022.

Predpol was developed in 2011, one of the first “predictive policing” services on the market. In 2020, a year before Worcester contracted ShotSpotter Connect, LAPD dropped PredPol citing the fact it didn’t do anything.

The April 2023 name change and rollout of a new “suite of technologies” were, according to Wired, “a rebranded and upgraded version of technology SoundThinking had acquired over the past four years.” Like all tech companies, SoundThinking is in the consolidation game. What Google and Facebook have done to the internet, they want to do for policing.

Saucier, or whoever wrote the report, has swallowed the Kool-Aid. Before ResourceRouter, the report reads, officers used to drive around “with no direction.”

Instead of officers driving around their patrol with no direction this technology directs them to a different location every eight hours. This takes place within all 20 routes in the city and is shown as a box on the officer's computer. The officers then visit these areas three times per shift for 15 min. These locations are citywide, not just within the area covered by ShotSpotter, and are within the officers' normal patrol area.
We now use the program to engage the community and deter crime by having our officers perform community engagement patrols within these predetermined areas.

This is what’s called “predictive policing” or “crime forecasting.” It’s the core function of ResourceRouter. The Electronic Freedom Foundation calls such technology a self-fulfilling prophecy. The police use past data, with all the biases present within it, to concentrate patrols in the high-volume areas. Those areas are now more likely to see more arrests, since the police are there more. ResourceRouter takes that data to “reflect that area as a hotbed of criminal activity.” Once it becomes a hotbed, the police can justify more surveillance. The EFF writes:

“This cycle results in further victimization of communities that are already mass policed—namely, communities of color, unhoused individuals, and immigrants—by using the cloak of scientific legitimacy and the supposedly unbiased nature of data.”

Back to the WPD report:

Officers are required to get out of their vehicles and conduct foot beats allowing them to meet community members. These community engagement patrols occur within the 20 routes citywide, suggested by ResourceRouter.

A 2023 Wired investigation found that these “directed patrol areas” also come with mandated tactics, including “patrol streets on foot,” as the WPD mentions, and “gather intel concerns from businesses,” which the WPD does not mention. What other tactics does this tech mandate officers to carry out? We don’t know. Also not mentioned by the WPD: the app monitors officer location—“‘Please return to directed patrol area,’ an affectless female voice warns,” if an officer strays from the assignment, according to Wired.

While such tech could be used to guard against overpolicing of certain areas, The Markup found that the opposite happened. “The people most likely to be affected by daily PredPol predictions were residents of public and subsidized housing, among the poorest residents.”

The WPD shared some raw bulk data spit out by the app: “In the last 6 months officers visited 11,172 locations for a total of 804,021 minutes.”

But didn’t share where those locations were, or what happened when officers went there. Instead, they offered a blanket reassurance:

Resource Router is not expected to result in arrests or major changes to our emergency response. Instead, it enhances our community policing foot beat efforts by directing officers to areas of their route where their presence is most likely to reassure residents and deter crime.

The statement “where their presence is most likely to reassure residents” is a loaded one. What does that mean? And who gets to say that with any honesty? Does the increased presence of police officers make you, the reader, feel more or less safe?

As SoundThinking gets exponentially larger, swallowing more competitors, municipal police departments stay roughly the same size. In the Wired piece, experts expressed concern that the consolidation of policing technology will produce a power imbalance on the “private” side of this “public-private partnership.”

“We haven’t had a conversation about what it means when a police chief is dependent on a tech company,” Ferguson says. “Are we OK ceding this much control over public safety?”

In Worcester, this is not an abstract concern. Last year, Saucier applied for a job as a ShotSpotter sales rep, after having spent years doing that job for free in our city. This is an idea I explored at length in “The police aren’t just customers, they’re salesmen.”

I’ve been trying to dig further into the worrying “revolving door” interplay between the company and our police department. I sent a records request for all emails between the company and the city over the past four years. The city responded saying there were too many emails to comply with my request and asked me to limit it. I sent another request last week for the past year and a half, as opposed to four. We’ll see if that works.

So much of Saucier’s writing on Shotspotter seems to be directly cribbed from the company’s marketing material, so the question of whether the company is dictating WPD policy is a legitimate one. Anyway, the report concludes:

The City of Worcester pays $48,750 per year for Resource Router and $524,120 per year for ShotSpotter.

That’s a total of $575,000 a year that the city takes from its residents and forks over to Silicon Valley. For what? For “the chief says it helps.” At the end of the day, that’s what underwrites the whole thing. If the chief supports it, your average municipal politician is going to get in line behind that support. Opposing the cops is, after all, politically fraught. Especially in a place like Worcester, where city elections are decided by a small group of mostly white and mostly middle class retirees.

The chief is trying to get a cushy job with the company while telling his community the company’s products help. His application’s strength rests on his ability to do so. And the majority of the political class takes his side so as to appear “moderate” and “sensible” to an electorate at best uninterested in racial and class inequities. This technology presents the perfect opportunity for reactionary panderers in elected positions to tacitly declare themselves “not woke.”

The whole thing is a scam that Worcester continues to fall for to an embarrassing degree. The company and the chief and the worst of our elected officials are the ones making out in the deal. Meanwhile it streamlines the violence wrought by over policing on our most oppressed communities.

I for one would rather see that $575,000 stuffed in a duffle bag and performatively thrown into a bonfire. It’d be one thing if it was money merely wasted, but this is money spent making things worse.

As things get worse, remember, the cops get more money. You could say they have a fiduciary responsibility to negative outcomes.

So onto question two:

Why are the police misrepresenting the data?

On Shotspotter’s main service, the gun detection mics, Saucier once again omitted the most crucial datapoint. And, once again, he misrepresented the data that he did provide. This is how he presented it:

In 2023 there were 116 confirmed gunshot incidents, indicating physical evidence was found on scene not including digital acoustic evidence, resulting in 70 ShotSpotter activations. 41 of these 70 incidents did not result in an immediate 911 call. The 70 ShotSpotter activations resulted in 12 arrests and the recovery of eight firearms in 2023.

All of the numbers are correct (presumably) but the verbs are off. The 116 overall confirmed gunshot incidents “resulted” in 70 ShotSpotter activations? What does that even mean?

What he’s not saying is there were 517 ShotSpotter activations (I break down the painstaking process of getting that number here.) Of those 517 activations, police were able to verify that someone fired a gun just 70 times. That’s 517 times they responded, with two beat cops and a supervisor, and investigated a scene. In 447 of those responses, they didn’t find anything. But they were nevertheless there, in some neighborhood, looking for trouble.

Graphic made for *Worcester Sucks* by a friend who will remain nameless.

Without the 517 number, the data Saucier provides is inscrutable. I do not think it’s an accident that he leaves it out. In most cases, he’s allowed to do so. We only know about the 517 number because the Human Rights Commission pressed him about it, and as far as I know I’m the only one to report it out.

There were 116 confirmed shootings. There were 70 in which Shotspotter played a role in the response. That means that there were 46 shootings that Shotspotter mics didn’t catch for whatever reason.

During the 70 times Shotspotter played some role in the police response, there were 12 arrests made and eight guns recovered.

This technology dispatched cops expecting gun fire to a specific location 517 times in 2023. Necessarily, those locations were in already overpoliced neighborhoods because that’s where the microphones were placed. More than 400 of those trips were dead ends. But the cops still went. They still took a look around. As Saucier describes:

When a ShotSpotter activation occurs, officers are responding to a dot on a map indicating within 25 meters (82 feet) of where the gunshot originated from. Dispatch has this same map and often directs officers in real time to the location of the dot. A supervisor is also dispatched to the scene per policy. Once on scene officers look for victims, witnesses, shell casings, property damage and others signs of a shot tired incident.

What happens on those dead end calls? What effect does it have on the people there? Are there any hostile interactions? Unrelated arrests? Do they pat people down? Do they make the situation tense? Does their presence turn up the heat?

I guess for lack of a better term I’ll call these possibilities “spin-off effects.” It’s a datapoint we really need and I don’t think any study out there has captured it. What else has happened on these 517 calls where cops go into a neighborhood anticipating a dangerous and violent situation? What happens when they can’t find evidence?

Getting that data would be a painstaking endeavor, even with a police department that complied with public records law. If you have an idea, get at me!

The true value of Shotspotter is to the bureaucracy. It produces new units of quantifiable “crime.” More crime, more bureaucracy. More justification for more money. It manufactures crime. Me and Chris Robarge collaborated to explore that idea at length in a post from March: “A Streamlining Solution for Crime Manufacturers.”

Thanks for reading! Again, I’m on vacation, so I’m blissfully unaware of any Worcester news past last Sunday.

Please subscribe if you appreciate this work!

venmo a tip :-)

paypal a tip :-)

And help spread the news! Word of mouth is the only type of piston in this outlet’s proverbial engine.

Share Worcester Sucks and I Love It

I’ll leave any of you hungry for more Worcester Sucks with this: